Chicken Swarm Optimization Algorithm Inspired by Chicken Behavior
1. A New Bio-inspired Algorithm Chicken Swarm Optimization
Xianbing Meng, Yu Liu, Xiaozhi Gao, and Hengzhen Zhang
Supervisor: Dr. Ahmed ElSawy
Presented by : Abdelrahman Alaa & Mohamed Wagih
2. Abstract
A new bio-inspired algorithm, Chicken Swarm Optimization (CSO), is proposed
for optimization applications. Mimicking the hierarchal order in the chicken
swarm and the behaviors of the chicken swarm, including roosters, hens and
chicks, CSO can efficiently extract the chickens’ swarm intelligence to optimize
problems. Experiments on twelve benchmark problems and a speed reducer
design were conducted to compare the performance of CSO with that of other
algorithms. The results show that CSO can achieve good optimization results in
terms of both optimization accuracy and robustness.
2
3. Agenda
Introduction
General Biology
Chicken Swarm optimization (CSO)
Movement of Chickens
Parametric Analysis
Validation and Comparison
Benchmark Problems Optimization
Speed Reducer Design
Discussion
References
3
5. 5
Introduction
Background
Chickens are kept as food source and Live together in flocks
Communicate using over 30 distinct sounds “clucks, cackles, chirps and cries”
including a lot of information “nesting, food discovery, mating and danger”
Learn through trial, error and previous experience
6. 6
Introduction
CSO mimics the hierarchal order in the chicken swarm
and the behaviors of the chicken swarm.
Hierarchal order? Behaviors of the chicken swarm?
8. 8
Introduction
Chicken swarm can be divided into several groups
each Group contains : 1 Rooster + many hens + many chicks
Competition between different chickens under specific order
9. 9
General Biology
Hierarchal order
A hierarchal order plays a significant role in the social lives of chickens
The preponderant chickens will dominate the weak
More dominant hens that remain near to the head roosters
The More submissive chicken stand at the periphery of the group
Removing or adding chickens from an existing group would causes
a temporary disruption to the social order until a specific hierarchal order
is established
10. 10
General Biology
Hierarchal order
The dominant individuals have priority for food access
Roosters may call their group-mates to eat first when they find food
Gracious behavior also exists in the hens when they raise their children.
However, this is not the case existing for individuals from different groups.
Roosters would emit a loud call when other chickens from a different group
invade their territory
11. 11
General Biology
Hierarchal order In General
The chicken’s behaviors vary with Gender
The head rooster would positively search for food, and fight with chickens
who invade the territory the group inhabits
The dominant chickens would be nearly consistent with the head roosters
to forage for food
The submissive ones, however, would reluctantly stand at the periphery of
the group to search for food. There exist competitions between different
chickens. As for the chicks, they search for the food around their mother
12. 12
Chicken Swarm Optimization (CSO)
Chickens’ behaviors rules
1- Chicken swarm divided into groups
each Group=a dominant , a couple of and
2- Group division methodology. All depends on the fitness values
Best fitness -> worst fitness and the rest are
Each would be the head rooster in a group
The hens randomly choose which group to live in.
The mother-child relationship is also randomly established.
13. 13
Chicken Swarm Optimization (CSO)
3- The hierarchal order, dominance relationship and mother-child relationship
in a group will remain unchanged. Only update every several (G) time steps
4- chicken follow their group-mate rooster to search for food While they
prevent the ones from eating their own food
Assume chickens would randomly steal the good food already found
by others.
The chicks search for food around their mother (hen)
The dominant individuals have advantage in competition for food.
RN “Roosters”, HN “hens”, CN “chicks” and MN “mother hens”
The best RN chickens would be assumed to be roosters
while the worst CN ones would be regarded as chicks
The rest are treated as hens
14. 14
Chicken Swarm Optimization (CSO)
All N virtual chickens, depicted by their positions
at time step t, search for food in a D-dimensional space.
In this work, the optimization problems are the minimal ones
Thus the best RN chickens -> the ones with RN minimal fitness values.
15. 15
Chicken Swarm Optimization (CSO)
Chickens Movement (Roosters)
The roosters with better fitness values have priority for food access than the ones
with worse fitness values.
For simplicity, Roosters with better fitness values can search for food in
a wider range of places than that of the roosters with worse fitness values
Randn (0, 𝝈 𝟐
) is a Gaussian distribution with mean 0 and standard deviation 𝝈 𝟐
𝜀, which is used to avoid zero-division-error, is the smallest constant in the computer
k, a rooster’s index, is randomly selected from the roosters group
f is the fitness value of the corresponding x.
16. 16
Chicken Swarm Optimization (CSO)
Chickens Movement (Hens)
Hens can follow their group-mate roosters to search for food.
They would also randomly steal the good food found by other chickens
Though they would be repressed by the other chickens.
The more dominant hens would have advantage in competing for food than the
more submissive ones
Rand is a uniform random number over [0, 1]
𝒓𝟏 ∈ [𝟏, … . . , 𝑵] index of the rooster, which is the ith hen’s group-mate
𝒓𝟐 ∈ [𝟏, … . . , 𝑵] index of the chicken (rooster or hen ), which is randomly chosen 𝒓𝟏 ≠ 𝒓𝟐
S1= exp(
𝑓 𝑖−𝑓𝑟1
|𝑓 𝑖|+𝜀
) (4) S2= exp(𝑓𝑟2 − 𝑓𝑖) (5)
17. 17
Chicken Swarm Optimization (CSO)
Chickens Movement (Hens)
Obviously 𝑓𝑖 > 𝑓𝑟1 , 𝑓𝑖 > 𝑓𝑟2 , thus S2 <1< S1
Assume S1=0,
then the ith hen would forage for food just followed by other chickens.
The bigger the difference of the two chickens’ fitness values
the smaller S2 and the bigger the gap between the two chickens’ positions is.
Thus the hens would not easily steal the food found by other chickens.
S1 and S2 formulas differs because there exist competitions in a group.
the fitness values of the chickens relative to the fitness value of the rooster
are simulated as the competitions between chickens in a group.
S2= exp(𝑓𝑟2 − 𝑓𝑖) (5)
18. 18
Chicken Swarm Optimization (CSO)
Chickens Movement (Hens)
Suppose S2=0,
then the ith hen would search for food in their own territory.
For the specific group, the rooster’s fitness value is unique.
Thus the smaller the ith hen’s fitness value, the nearer S1 approximates to 1
and the smaller the gap between the positions of the ith hen and its group-mate rooster
Hence the more dominant hens
would be more likely than the more submissive ones to eat
19. 19
Chicken Swarm Optimization (CSO)
Chickens Movement (Chicks)
The chicks move around their mother to forage for food. This is formulated below
𝒙 𝒎,𝒋
𝒕
stands for the position of the ith chick’s mother (𝒎 ∈ [𝟏, 𝑵])
𝑭𝑳(𝑭𝑳 ∈ 𝟎, 𝟐 ) parameter indicates that the chick would follow its mother to forage for food
Consider the individual differences, the FL of each chick would randomly choose between 0 and 2
21. 21
Chicken Swarm Optimization (CSO)
Algorithm
• Individual of chicken swarm population are initialized by using the
following formula
• With lb and ub are lower bound and upper bound of the search
space.
22. 22
Chicken Swarm Optimization (CSO)
Parametric Analysis
There exist six parameters in CSO.
HN would be bigger than RN -> keeping hens is more beneficial for human because only
hens can lay eggs, which can also be the source of food
HN is also bigger than MN -> not all hens would hatch their eggs simultaneously
Though each hen can raise more than one chick, we assume the population of adult chickens
would surpass that of the chicks, CN
As for G, it should be set at an appropriate value, which is problem-based.
If G is very big-> it's not conducive for the algorithm to converge to the global optimal quickly.
If G is very small, the algorithm may trap into local optimal.
After the preliminary test, G ∈ [2,20] may achieve good results for most problem.
In practice, FL ∈ [0.4, 1] usually perform well.
The formula of the chick’s movement can be associated with the corresponding part in DE
If we set RN and MN at 0, thus CSO essentially becomes the basic mutation scheme of DE.
23. 23
Benchmark Problems Optimization
Twelve popular benchmark problems (shown in Table 1) are used to verify the
performance of the CSO compared with that of PSO, DE and BA.
The statistical results have been obtained, based on 100 independent trials, in all
the case studies. The number of iterations is 1,000 in each trial.
For a fair comparison, all of the common parameters of these methods, such as the
population size, dimensions and maximum number of generations, are set to be the
same. The related parameters of these algorithms are showed at Table 2
Validation and Comparison
26. 26
Benchmark Problems Optimization
The superiority of CSO over PSO, BA and DE should be the case.
If we set RN = CN = 0, and let S1, S2 be the parameters like c1 and c2 in PSO, thus
CSO will be similar to the standard PSO. Hence CSO can inherit many advantages
of PSO and DE.
Moreover, the chickens’ swarm intelligence can be efficiently extracted in CSO.
Given the diverse laws of the chickens' motions and cooperation between the
multigroups, the search space can be efficiently explored.
Under the specific hierarchal order, the whole chicken swarm may behave like a
team to forage for food, which can be associated with the objective problems to
be optimized. All of these merits enhance the performance of CSO.
Validation and Comparison
29. 29
Speed Reducer Design – design Gearbox
Which can be rotated at its most efficient speed.
The gearbox is described by
The face width 𝑏(𝑋1)
Module of teeth 𝑚(𝑋2)
Number of teeth in the pinion 𝑍 𝑋3
Length of the first shaft between bearings ℎ1 𝑋4
Length of the second shaft between bearings ℎ2 𝑋5
Diameter of the first shaft 𝑑1 𝑋6
Diameter of the first shaft 𝑑2 𝑋7
Speed Reducer Design optimization is to minimize its total weight, subject
to constraints on bending stress of the gear teeth, surface stress, transverse
deflections of the shafts, and stresses in the shafts
Validation and Comparison
30. 30
Application - Speed Reducer Design – design Gearbox
This problem can be formulated as follows
Validation and Comparison
31. 31
Validation and Comparison
Speed Reducer Design – Design Gearbox
the results achieved by CSO and other algorithms.
CSO’s results outperform all the results achieved by other methods
in terms of both optimization accuracy and robustness
which indicates that the solution is feasible.
32. 32
Discussion
The performance of CSO is compared with that of the PSO, DE and BA on twelve
benchmark problems.
Experiments show that CSO outperforms the PSO, DE and BA in terms of both
optimization accuracy and robustness.
Moreover, CSO can efficiently solve the speed reducer design, which endues the CSO
with a promising prospect of further studying.
One of the reasons that CSO has very promising performance is that CSO inherits major
advantages of many algorithms. PSO and the mutation scheme of DE are the special
cases of the CSO under appropriate simplifications.
What is more significant for the superiority of the CSO is that the chickens’ swarm
intelligence can be efficiently extracted to optimize problems.
The chickens' diverse movements can be conducive for the algorithm to strike a good
balance between the randomness and determinacy for finding the optima.
33. 33
Discussion
The whole chicken swarm consists of several groups, namely multi-swarm. Through
integration of the hierarchal order, chickens of the different groups may behave as a team
and coordinate themselves to forage for food. Thus CSO can behave intelligently to
optimize problems efficiently.
The innovation in this paper not only lies in efficiently extracting the chickens’ swarm
intelligence to optimize problems, but also making CSO innate multi-swarm method.
Multi-swarm technique is usually used to enhance performance of the population-
based algorithm. As an innate multi-swarm algorithm, various multi-swarm techniques can
be used to develop the different variants of CSO. Thus CSO has good extensibility.
Moreover, from the parametric analysis, the population of the hens is the biggest in the
swarm. Thus the performance of CSO largely depends on how the hens’ swarm
intelligence can be extracted to optimize problems.
34. 34
Discussion
The motion of the hens can be adaptively controlled according to the fitness value of
the problem itself.
With the dynamical hierarchal order, the hens swarm can be updated. Hence CSO has
the self adaptive ability to solve the optimization problems.
More comprehensive analyses on the CSO are still need to be investigated in the future.
Moreover, we can consider there exist several roosters in a group and dynamically
adjust the population of the hens and chicks in each group.
It’s also significant to tune the related parameters for enhancing the algorithm
performance, and design the variants of the CSO to solve many optimization applications.
35. 35
References
References
1. Yang, X.S.: Bat algorithm: literature review and applications. International Journal of Bioinspired Computation 5(3), 141–149 (2013)
2. Das, S., Suganthan, P.N.: Differential evolution: A survey of the state-of-the-art. IEEE Transactions on Evolutionary Computation 15(1), 4–31 (2011)
3. Jordehi, A.R., Jasni, J.: Parameter selection in particle swarm optimization: A survey. Journal of Experimental & Theoretical Artificial Intelligence 25(4), 527–
542 (2013)
4. Gandomi, A.H., Alavi, A.H.: Krill herd: A new bio-inspired optimization algorithm. Communications in Nonlinear Science and Numerical Simulation 17,
4831–4845 (2012)
5. Cuevas, E., Cienfuegos, M., Zaldivar, D., Cisneros, M.: A swarm optimization algorithm inspired in the behavior of the social-spider. Expert Systems with
Applications 40, 6374–6384 (2013)
6. Smith, C.L., Zielinski, S.L.: The Startling Intelligence of the Common Chicken. Scientific American 310(2) (2014)
7. Grillo, R.: Chicken Behavior: An Overview of Recent Science, http://freefromharm.org/chicken-behavior-an-overview-ofrecent-science
8. Chicken, http://en.wikipedia.org/wiki/Chicken
9. Tan, Y., Li, J.ZYang, X.S.: Nature-inspired optimization algorithm. Elsevier (2014)
10. Robert, R., Mostafa, A.: Embedding a social fabric component into cultural algorithms toolkit for an enhanced knowledge-driven engineering optimization.
International Journal of Intelligent Computing and Cybernetic 1(4), 563–597 (2008)
11. Mezura, M.E., Hernandez, O.B.: Modified bacterial foraging optimization for engineering design. In: Proceedings of the Artificial Neural Networks in
Engineering Conference, vol. 19, pp. 357–364. Intelligent Engineering Systems Through Artificial Neural Networks (2009)
12. Akay, B., Karaboga, D.: Artificial bee colony algorithm for large-scale problems and engineering design optimization. Journal of Intelligent Manufacturing
23(4), 1001–1014 (2012)
13. Gandomi, A.H., Yang, X.S., Alavi, A.H.: Cuckoo search algorithm: A metaheuristic approach to solve structural optimization problems. Engineering with
Computers 29, 17–35 (2013)